Presented by: Michal Maly, Photoneo
The most modern and innovative approach to advancing the world of automation to new heights is the implementation of artificial intelligence into industrial processes, making them faster and more effective.
On the one hand, AI is enabling completely new applications. With classical computer vision algorithms, it might be difficult to handle items of irregular shapes, forms, or colors - for example, boxes, parcels, and organic materials such as fruits. This is especially true if the conditions vary. AI offers great flexibility, eliminating manual fine-tuning of algorithms.
On the other hand, AI can also enhance existing approaches. However, how to use machine learning must be carefully considered. In many recipe-oriented inspection tasks, such as the checking of strictly pre-defined dimensions, it is impractical to assume that machine learning will learn or deduce the recipe. The role of machine learning should be kept to finding objects (edges, parts, etc.) or detecting vaguely described faults.
Maly will discuss the general concepts of AI, how they are applied to modern automation processes, and the challenges that AI may pose. One technical challenge, for instance, is the trade-off between cost, computation time and latency (especially in embedded devices), and flexibility. On the business side is a communication challenge - helping customers to understand the role of AI, especially how it is trained and how it works.
Another limit is that machine learning is often used or implemented in opposition to precisely defined measures, which means that it is example-driven rather than specification-driven. Therefore, in critical applications, it needs to be supplemented by additional monitoring, system checks, or post-processing algorithms.
About the presenter
Michal Maly, Ph.D., is co-founder and director of AI at Photoneo. He received his doctorate in computer science and artificial intelligence from Comenius University in Bratislava, Slovakia, in the Faculty of Mathematics, Physics, and Informatics in 2013. He is an expert in the construction of reinforcement learning agents based on rationality. He also worked in the area of computer security and published an innovative solution for secure distributed computing. His vision is to improve computer understanding of the 3D environment and enhance comprehension of real-world objects.